Skip to main content

Self-reflective symbolic AI: real-time phase-transition detection and native self-reflection loop with mirror-based cognition

Project description

aeon-ai

Self-reflective symbolic AI — real-time phase-transition detection & native self-reflection loop — GenesisAeon v0.2.0

PyPI version Python License: MIT DOI Tests

”A system that thinks — a memory that sings.”

aeon-ai is the core Python library of the GenesisAeon project: a fractal, self-reflective AI architecture that integrates fieldtheory Lagrangian dynamics, mirror-machine cognition, symbolic sigil activation, and CREP quality evaluation into a unified pipeline.


Theoretical Foundation

Fieldtheory Lagrangian — AeonLayer

The core weighting dynamics are governed by the fieldtheory Lagrangian:

$$L(S_A, S_V, \delta, t) = \frac{S_A \cdot S_V}{S_A + S_V} - \frac{1 + \delta}{t^2}$$

Symbol Meaning
$S_A$ Auditory (abstract) signal amplitude
$S_V$ Visual (visceral) signal amplitude
$\delta$ Deformation / curvature parameter
$t$ Time step ($t > 0$)

The first term is the harmonic mean of the two signal amplitudes, encoding inter-channel resonance. The second term is a temporal penalty that decays as $t^{-2}$, reflecting the dissipation of symbolic tension over time.

Analytical gradients:

$$\frac{\partial L}{\partial S_A} = \frac{S_V^2}{(S_A + S_V)^2}, \quad \frac{\partial L}{\partial S_V} = \frac{S_A^2}{(S_A + S_V)^2}, \quad \frac{\partial L}{\partial t} = \frac{2(1+\delta)}{t^3}$$


CREP Quality Metric — CREPEvaluator

CREP is the four-dimensional symbolic quality metric from the unified-mandala stack:

$$\text{CREP} = \frac{4}{\dfrac{1}{C} + \dfrac{1}{R} + \dfrac{1}{E} + \dfrac{1}{P}}$$

Dimension Symbol Description
Coherence $C$ Logical and structural consistency
Resonance $R$ Harmonic alignment between signal components
Emergence $E$ Novelty and self-organisational complexity
Poetics $P$ Aesthetic and symbolic richness

Each dimension is a float in $[0, 1]$; the combined score is their harmonic mean.

A weighted variant is also available:

$$\text{CREP}w = \sum{i \in {C,R,E,P}} w_i \cdot \text{dim}_i, \quad \sum w_i = 1$$


UTAC-Logistic — MirrorCore

The Universal Transformation and Adaptation Coefficient governs the INTEGRATE phase of the Mirror Machine:

$$\text{UTAC}(x) = \frac{L}{1 + e^{-k(x - x_0)}}$$

Symbol Meaning
$L$ Carrying capacity (saturation ceiling)
$k$ Growth rate / sigmoid steepness
$x_0$ Inflection / midpoint

Mirror Machine Pipeline

The self-reflection loop processes every input through four canonical phases:

INIT → REFLECT → INTEGRATE → EMIT
  1. INIT — Receive raw signal, compute initial entropy
  2. REFLECT — Apply $n$-deep recursive tanh-mirror transformation
  3. INTEGRATE — Merge with context memory via UTAC-Logistic
  4. EMIT — Produce observable output, update persistent memory

Phase-Transition Detection — PhaseDetector (v0.2.0)

Real-time detection of Mirror Machine phase transitions using a UTAC-Logistic trigger:

$$\Phi_{\text{trigger}}(H) = \frac{L}{1 + e^{-k(H - H_0)}}$$

Symbol Meaning
$H$ Current Shannon entropy
$H_0$ Pivot entropy threshold (default 0.37)
$k$ UTAC growth rate
$L$ Carrying capacity

A collapse is detected when consecutive output values converge:

$$\Delta_{\text{collapse}} = \left| x_n - x_{n-1} \right| < \epsilon_{\text{stab}}$$

Three transition types are emitted: FORWARD, COLLAPSE, UTAC_TRIGGER, FORCED.


Native Self-Reflection Loop — SelfReflector (v0.2.0)

A closed-loop iterative engine coupling CREP score, Lagrangian gradient, and Sigillin bridges over at most 7 iterations with convergence check:

$$S_A^{(i+1)} = S_A^{(i)} + \eta \cdot \frac{\partial L}{\partial S_A}$$

$$S_V^{(i+1)} = S_V^{(i)} + \eta \cdot \frac{\partial L}{\partial S_V}$$

$$\text{convergence: } \left|\text{CREP}^{(i)} - \text{CREP}^{(i-1)}\right| < \varepsilon$$

Where $\eta$ is the UTAC-modulated step size and CREP is evaluated at each iteration.


Architecture Overview

┌────────────────────────────────────────────────────────────────┐
│                        Orchestrator v0.2.0                      │
│                                                                  │
│  FieldBridge ──▶ CosmicMoment                                   │
│       │                │                                         │
│       ▼                ▼                                         │
│  AeonLayer(δ, t) ──▶ Lagrangian L                               │
│       │                                                          │
│       ▼                                                          │
│  MirrorCore ──▶ INIT → REFLECT → INTEGRATE → EMIT               │
│       │                │                                         │
│       ▼                ▼                                         │
│  PhaseDetector ──▶ PhaseTransitionEvent[]  (v0.2.0)             │
│       │                                                          │
│       ▼                                                          │
│  CREPEvaluator ──▶ CREPScore(C, R, E, P)                        │
│       │                                                          │
│       ▼                                                          │
│  SigillinBridge ──▶ { sigil_id: activation_score }              │
└────────────────────────────────────────────────────────────────┘

SelfReflector (v0.2.0) — closed-loop coupling (max 7 iterations):
  AeonLayer → MirrorCore → CREPEvaluator ← SigillinBridge
       ↑_______________gradient update__________________|

Installation

pip install aeon-ai

With the full GenesisAeon stack (mirror-machine, fieldtheory, mandala-visualizer, …):

pip install 'aeon-ai[stack]'

Quick Start

from aeon_ai.agents import Orchestrator

# Full pipeline via Orchestrator
orch = Orchestrator(delta=0.1, mirror_depth=3)
result = orch.run(
    s_a=0.8,
    s_v=0.6,
    sigil_text=mirror aeon genesis,
)

print(fLagrangian L     = {result.lagrangian_out:.4f})
print(fReflection out   = {result.reflection.output_val:.4f})
print(fFinal output     = {result.final_output:.4f})
print(fCREP score       = {result.crep_score.score:.4f})
print(fActive sigils    = {result.sigil_activations})

Individual components

from aeon_ai import AeonLayer, lagrangian

# Direct Lagrangian computation
L = lagrangian(s_a=0.8, s_v=0.6, delta=0.1, t=1.0)

# AeonLayer (with optional advanced_weighting_systems base)
layer = AeonLayer.from_advanced_weighting_systems(delta=0.1)
output = layer.forward(s_a=0.8, s_v=0.6, t=1.0)
grad   = layer.gradient(s_a=0.8, s_v=0.6, t=1.0)
from aeon_ai import MirrorCore

core = MirrorCore(depth=3, utac_growth=2.0)
state = core.reflect(0.7, entropy=0.4)
print(state.output_val, state.entropy)
from aeon_ai import CREPEvaluator

ev = CREPEvaluator()
score = ev.evaluate(signal=[0.3, 0.7, 0.5, 0.9], text=aeon of mirrors)
print(score)  # CREPScore(C=..., R=..., E=..., P=..., score=...)
from aeon_ai import SigillinBridge

bridge = SigillinBridge()
activations = bridge.activate(the mirror reflects the genesis of aeon)
expanded    = bridge.poetic_expansion(aeon rises)

CLI

# Basic reflection run
aeon reflect --models trans,cnn --sigil “mirror aeon genesis” --entropy 0.4

# With custom signal parameters
aeon reflect --s-a 0.9 --s-v 0.5 --delta 0.2 --time-step 2.0

# With mandala visualisation + sonification (requires [stack])
aeon reflect --sigil “origin seed” --entropy 0.6 --visualize

# Machine-readable JSON output
aeon reflect --sigil “mirror” --json

# List all registered sigils
aeon sigils

# Package and stack status
aeon info

Module Reference

Module Class / Function Description
aeon_ai.aeon_layer AeonLayer Lagrangian weighting layer
aeon_ai.aeon_layer lagrangian() Fieldtheory Lagrangian function
aeon_ai.mirror_core MirrorCore Self-reflection pipeline
aeon_ai.mirror_core utac_logistic() UTAC-Logistic function
aeon_ai.crep_evaluator CREPEvaluator CREP quality scorer
aeon_ai.sigillin_bridge SigillinBridge Sigil activation engine
aeon_ai.field_bridge FieldBridge Cosmic-moment modulation
aeon_ai.agents Orchestrator Full pipeline coordinator
aeon_ai.phase_detector PhaseDetector Real-time phase-transition detector (v0.2.0)
aeon_ai.phase_detector detect_phases_from_core() One-shot trace analysis (v0.2.0)
aeon_ai.phase_detector entropy_phase_label() Entropy → phase label (v0.2.0)
aeon_ai.self_reflection SelfReflector Closed-loop self-reflection engine (v0.2.0)
aeon_ai.self_reflection ReflectionLoopResult Loop result record (v0.2.0)

Full API documentation: genesisaeon.github.io/aeon-ai


Development

git clone https://github.com/GenesisAeon/aeon-ai
cd aeon-ai
pip install -e '.[dev]'

# Tests (99%+ coverage required)
pytest

# Linting
ruff check src tests
ruff format --check src tests

# Docs
mkdocs serve

GenesisAeon Stack

aeon-ai is designed to interoperate with the full GenesisAeon ecosystem:

Package Role
advanced-weighting-systems Base AeonLayer weights
fieldtheory Cosmological field dynamics
mirror-machine Deep mirror-pass kernels
entropy-governance Adaptive entropy regulation
sigillin Extended sigil registry
utac-core UTAC kernel implementations
mandala-visualizer Mandala network rendering
cosmic-web Field sonification

All packages are optional; aeon-ai operates standalone with its own implementations.


Citation

If you use aeon-ai in academic work, please cite:

@software{aeon_ai_2025,
  author    = {GenesisAeon},
  title     = {aeon-ai: Real-time phase-transition detection and native self-reflection loop},
  version   = {0.2.0},
  year      = {2026},
  doi       = {10.5281/zenodo.19132293},
  url       = {https://github.com/GenesisAeon/aeon-ai}
}

DOI: 10.5281/zenodo.19132293


License

MIT MIT — see LICENSE.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

aeon_ai-0.3.1.tar.gz (22.8 MB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

aeon_ai-0.3.1-py3-none-any.whl (39.1 kB view details)

Uploaded Python 3

File details

Details for the file aeon_ai-0.3.1.tar.gz.

File metadata

  • Download URL: aeon_ai-0.3.1.tar.gz
  • Upload date:
  • Size: 22.8 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for aeon_ai-0.3.1.tar.gz
Algorithm Hash digest
SHA256 de8ac80edc750a9c9f4989aa4b01484ed622827e4681548bb2aa8af31606e569
MD5 8947c35dbc7f9af005c69a1aa762d93a
BLAKE2b-256 7dc2097f51a854181c6430fb93b967c641efed3312e113e2be5bbf9bcca29085

See more details on using hashes here.

Provenance

The following attestation bundles were made for aeon_ai-0.3.1.tar.gz:

Publisher: release.yml on GenesisAeon/aeon-ai

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file aeon_ai-0.3.1-py3-none-any.whl.

File metadata

  • Download URL: aeon_ai-0.3.1-py3-none-any.whl
  • Upload date:
  • Size: 39.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for aeon_ai-0.3.1-py3-none-any.whl
Algorithm Hash digest
SHA256 7a6e0d25d44367de13183195b1bdeb7638699160d32c83a4bd946414701f4648
MD5 29f7cd6054ba8c203a5176e6069c8a54
BLAKE2b-256 a8972c64f0db7198f9a2536b6e9bf21849460db979522896bf54810ead8b47af

See more details on using hashes here.

Provenance

The following attestation bundles were made for aeon_ai-0.3.1-py3-none-any.whl:

Publisher: release.yml on GenesisAeon/aeon-ai

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page